The global economy is entering an unusual phase: artificial intelligence is supporting growth even as it intensifies inequality between countries, strains energy systems and raises questions about whether today’s investment boom can deliver tomorrow’s profits.
The International Monetary Fund’s October outlook projects global growth of 3.1% in 2026, according to reporting on the fund’s latest assessment. But the headline figure masks a sharper divergence. Taiwan, at the centre of the advanced-chip supply chain, is projected to grow by 10.7% this year, while countries outside the technology and commodity chains face higher energy bills and weaker demand.
The IMF describes the result as a “split global economy”: technology-related exports and favourable terms of trade are cushioning some economies, while others are absorbing the costs of geopolitical conflict and imported fuel.
Why AI is becoming a macroeconomic force
The first driver is investment. Data centres, advanced semiconductors, electricity networks and cooling systems require enormous capital spending. That spending is now large enough to affect national growth rates, trade balances and financial markets, rather than remaining a narrow technology-sector story.
Taiwan illustrates the advantage. Its chip manufacturers are supplying the processors needed for cloud computing and AI systems, giving the island a surge in exports and investment. The IMF expects its growth to slow to 4.3% in 2027, suggesting that even the strongest beneficiaries may face a normalisation once construction and inventory cycles weaken.
The second driver is energy. AI infrastructure consumes substantial electricity, and reliable power has become a competitive asset. Countries able to provide cheap, stable electricity—and the grids, permitting systems and water resources required by large data centres—can attract investment. Those dependent on imported oil and gas are more exposed when conflict pushes prices higher.
Recent market reporting put oil at $117 a barrel amid fears of an escalation in hostilities involving Iran. Even if the price shock proves temporary, it highlights the vulnerability of economies that must pay more for energy while receiving little of the investment upside from AI.
The investment boom carries financial risks
AI spending is also creating a debate over valuations and debt. Deloitte reported that investors were increasingly concerned that revenue from AI companies may not be sufficient to service the borrowing used to build infrastructure. It cited rising credit-default-swap prices and bond yields for AI-related borrowers as signs of growing caution.
That does not prove the technology is a bubble. Supporters argue that current spending is laying the foundation for productivity gains across manufacturing, services, medicine and logistics. The IMF likewise sees AI investment as supporting near-term activity. If businesses eventually use these systems to produce more with fewer resources, the initial capital costs could be justified.
The opposing case is that infrastructure is being built ahead of proven demand. Gartner has estimated that 80% of “neocloud” providers could cease operating independently by 2030, a forecast reflecting the gap between capital expenditure and likely returns. A correction would not affect all economies equally: semiconductor exporters might face falling orders, while highly indebted companies and utilities could be left with expensive, underused assets.
What governments can do
The policy challenge is to capture productivity gains without allowing the boom to become a source of financial instability or social dislocation. Governments are competing to attract data centres and chip plants, often through subsidies, tax incentives and looser planning rules. Those measures may create jobs and strengthen strategic supply chains, but they can also transfer risk to taxpayers if projects fail to earn commercial returns.
Energy policy is central. Expanding grids, accelerating renewable generation, maintaining storage and improving efficiency can make AI growth less dependent on volatile fossil-fuel markets. Yet rapid construction can collide with local concerns over land, water and electricity prices. A strategy focused only on attracting facilities may deepen pressure on communities without guaranteeing broad economic benefits.
There is also a distributional question inside countries. AI-related gains are likely to accrue first to owners of capital, specialised workers and regions with strong infrastructure. Policymakers therefore face pressure to invest in education, competition and worker transition programmes, while preventing dominant firms from controlling essential computing capacity.
What comes next
The immediate outlook is likely to remain uneven. AI investment may keep supporting growth in chip-producing economies, while energy shocks and high interest rates weigh on importers. Markets will watch whether technology companies can turn enormous capital spending into recurring cash flow, and whether electricity demand can be met without reversing emissions reductions.
The most consequential test is not whether AI expands, but who captures its gains and who pays for its infrastructure. The IMF’s forecast shows that the technology boom can lift global growth while simultaneously widening national divides. The next phase will be shaped by choices over energy, public subsidies, financial regulation and access to the skills needed to participate.